Download PDF

Agricultural Economics, 1 (1987) 195-207
Elsevier Science Publishers B.V., Amsterdam- Printed in The Netherlands
195
On Why Agriculture Declines with Economic
Growth
Kym Anderson
Department of Economics, University of Adelaide, Adelaide S.A. 500 1 (Australia)
(Accepted 15 April 1987)
Abstract
Anderson, K., 1987. On why agriculture declines with economic growth. Agric.
Econ., 1: 195-207.
When economic growth is characterised by a slow rise in the demand for food
and rapid growth in farm relative to non-farm productivity, it is understandable
that agriculture in a closed economy declines in relative terms as that economy
develops. But why should agriculture decline in virtually all open growing economies as well, including those able to retain a comparative advantage in agricultural products? A key part of the answer is that the demand for non-tradable
goods tends to be income elastic, so resources are diverted to their production
.
.
even m open economLes.
Introduction
One of the dominant changes that characterises a growing economy is the
proportionate decline in the agricultural sector. This phenomenon is commonly attributed to two facts: the slower rise in the demand for food as compared with other goods and services, and the rapid development of new farm
technologies which lead to expanding food supplies per hectare and per worker
(Schultz, 1945; Kuznets, 1966; Johnson, 1973).
It is true that such conditions could lead to a relative decline of agriculture
globally, or in a smaller closed economy. But why does agriculture decline in
virtually all open growing economies? Might not some countries with rapid
growth in agricultural relative to non-agricultural productivity improve their
comparative advantage in agriculture sufficiently to prevent their agricultural
sectors from declining in relative terms?
The purpose of this paper is to use standard theory and empirical realities
0169-5150/87/$03.50
© 1987 Elsevier Science Publishers B.V.
196
TABLE 1
The declining importance of agriculture, 1960 to 1981
GDP
Employment
Exports
Relative labour
productivity
in agriculture
Low-income countries
1960
1981
48
37
77
73
70
31
0.62
0.51
Lower middle-income countries
1960
1981
36
22
71
54
76
39
0.51
0.41
Upper middle-income countries
1960
1981
18
10
49
30
46
18
0.37
0.33
6
3
18
6
23
15
0.32
0.52
Share of agriculture (%) in
Industrial market economies
1960
1981
"Agriculture's share of GDP divided by its share of employment.
Source: World Bank (1983, 1984, 1985).
to explain why agriculture tends to decline even for economies able to retain a
strong comparative advantage in agriculture.
The evidence of decline
The relative decline of agriculture is clear from both cross-sectional and
time-series data for the major country groupings shown in Table 1. The negative relationships between agriculture's shares of gross domestic product
( G D PSH), employment ( EMPSH) and exports ( EXPSH) on the one hand,
and income per capita ( YPC) on the other, are very significant statistically.
These shares are also negatively associated with population density per unit
of agricultural land ( PD), although significantly so only for EXPSH. This is
clear from the following OLS regression equation estimates, which are based
on World Bank data for 1981 for 70 countries with populations in excess of 1
million ( t-values in parentheses):
GDPSH = 87
9.2ln YPC
( 6.7)
EMPSH= 179
18.5ln YPC
(16.6)
197
A'"
Food
F'
F"
0
M
B
M'
B'" M"
B'
B"
Manufactures
Fig. 1. The world (or closed) economy.
EXPSH
= 152
-
9.5ln YPC
( 5.1)
8.5ln PD
( 4.7)
The measured rates of decline in agriculture's importance would be even faster
were it not for the fact that low-income countries tend to underprice agricultural products while high-income countries tend to overprice them.
Reasons for agriculture's relative decline
GOP share
To understand why agriculture declines, it is simplest to proceed in stages,
beginning with the GD P share. Consider first the global (or a closed) economy
producing with available resources only two final goods, food (F) from the
agricultural sector and non-food manufactures (M) from the rest of the economy. If AB in Fig. 1 represents the production possibility frontier for the economy in the first period and U a consumption indifference curve, then tangency
point E will be the point of equilibrium where supply equals demand for both
goods. The equilibrium price ratio is the slope of price line 1.
Suppose that with economic growth the production possibility curve moves
out equi-proportionately. Then because the income elasticity of demand for
food is less than one (Engel's Law), the new equilibrium pointE' will be to
the southeast of the ray OE extended and the new price line 2 will be steeper
198
than price line 1 (assuming, as is reasonable, that the shift in the production
possibility curve is not solely due to population growth, so that per capita income
rises). That is, the ratio of food to non-food production falls and so too does
the ratio of food to non-food prices. Hence agriculture's share of GDP must
fall as a result of this pattern of economic growth (where GDP is the summed
product of price and quantity for the two sectors in this simple model with no
intermediate inputs).
If non-food manufactured goods are the numeraire and pis the price of food
in terms of manufactures, then from Fig. 1 gross domestic product in this model
increases from Fp + M to F'p' + M', where ' indicates the new quantities and
price ratio. Since M/F < M' /F' andp> p', agriculture's share ofGDP decreases
from Fp/ (Fp+ M) to F'p' / (F'p' + M') or, equivalently, from 1/ (1 + M/Fp)
to 1/ (1 + M' /F'p').
Under what circumstances would the GDP share of this economy's food sector expand with economic growth? There are two extreme possibilities, both
shown in Fig. 1. One involves productivity growth being heavily biased in favour
of non-food, as with a shift to E"; the other involves the opposite extreme
condition, as at E "'. In both cases the sectoral productivity bias has to be sufficiently extreme for the effect on the price ratio to more than offset the opposite effect on relative quantities ofF and M. In terms of Fig. 1, the pointE"
and the slope of price line 3 would need to be such that M/Fp> M" /F"p".
Since M/F < M' /F' < M" /F", a very large increase in the price of food relative
to non-food- that is, a very skewed shift in the production possibility curve
in favour of non-food- would be required to prevent agriculture's share of
GDP from falling as this economy grows. Even if there were to be no productivity growth in agriculture, it is still possible that the skewed effect of income
growth on demand could outweigh the skewed effect of productivity growth on
supply so that the price of food relative to non-food, and hence agriculture's
share of GDP, falls in this growing global (or closed) economy.
The other extreme case, atE"', seems even less likely: if growth were to be
sufficiently biased in favour of the food sector to ensure a new equilibrium at
E 0 on the ray OE extended, for example, so that the relative quantities ofF and
M produced were the same as at E, the lower relative price of food at E 0 would
still ensure a drop in the food sector's share of GDP. That is, E "' has to be even
further to the northwest corner of Fig. 1 than E 0 •
Two reasons can be given as to why an outcome such as E"' is unlikely. One
is that since agriculture typically is the more labour-intensive sector of a developing country, the accumulation of capital per worker as economic growth proceeds will tend to encourage expansion ofthe more capital-intensive sector at
the expense of the more labour-intensive sector. This is a standard trade theorem due to Rybczynski (1955). The second reason is based on an empirical
observation. While the empirical evidence on the long-run trend in the price
of primary products relative to manufactures is still subject to debate, partie-
199
200
150
100
75
1900
1910 1920 1930 1940 1950 1960 1970 1960
Fig. 2. Real international prices of agricultural products, 1900-1985 (1979-81 = 100). Weighted
average of international prices in US dollars for grains, livestock products, sugar, beverages, vegetable oils, bananas, fibres, rubber, tobacco, hides and skins, and timber, deflated by the qualityadjusted US producer price index for industrial goods, with weights based on the share of each
commodity in the value of world trade in 1979-81. When the logarithm of this relative price index
(P) is regressed against time ( T), the equation obtained is P= 0.25-0.0051 T UP= 0.48; tvalue= -8.8). (Compiled by the author from data provided in Grilli and Yang (1986) and supplemented with updated data from World Bank files for which the author is indebted to M.C.
Yang.)
A,
Food
A"
A
I
I
F'
------+---1
I
I
I
0
B"
M
Manufactures
Fig. 3. A small, open economy.
ularly on the extent to which quality adjustments have been made to available
price indexes of manufactured goods, the weight of evidence seems to support
the view that agricultural prices have declined relative to industrial product
prices (see, for example, Spraos, 1980; Sapsford, 1985; Grilli and Yang, 1986).
One comprehensive series, developed by the World Bank for the period
1900-1985, is shown in Fig. 2. The rate of decline is slightly faster if beverages
are excluded, and slightly slower if non-food agriculture is excluded, but in all
200
cases the estimated coefficients for the time variable in the log-linear regression equations are significantly less than zero at the 0.1% level. Together these
observations suggest that outcomes to the northwest of E 0 such as E"' in Fig.
1 are most unlikely, and so will be ignored.
The above model is appropriate for the world economy as a whole, and for a
closed national economy. Consider now a small country in an open global economy, with again only two sectors producing the tradable products F and M.
Suppose that economic growth in the rest of the world is not strongly biased
against agriculture, so the international price of food relative to manufactures
is falling as Fig. 2 suggests. If there is no growth in this small country then the
decline internationally in the relative price of food from p to p' will result in a
decline in the share of agriculture in this economy's GDP. This is clear from
the movement from E to E' in Fig. 3, since both the price and the quantity of
food relative to manufactures fall.
What if this small economy is also growing? If its growth is such that the
production possibility curve moves out equi-proportionately, then agriculture's share of GDP must still decline. This is because in Fig. 3 F" /M" would
then equal F' /M' ( <F/M) and, since p' is still the ruling international and
domestic price ratio, it follows that 1/ (1 + M/Fp) > 1 (1 + M" /F"p'). Only if
the shift in this country's production possibility curve is skewed in favour of
agriculture, such as from AB to A"' B"', would it be possible for the increase in
production of food relative to manufactures to more than offset the decrease
in the relative price of food.
Three points are thus worth stressing so far. First, in a growing world economy (or a closed national economy) agriculture's share of total product is
likely to decline because the income elasticity of demand for food is less than
one. Only if the expansion in production possibilities is heavily biased towards
non-agriculture is it possible that the price of food relative to other goods will
rise - and even then this price change may not be sufficient to offset the
opposite change in relative production levels and thereby increase agriculture's
share of total product. Second, in a small open economy faced with a decline
in the relative price of food because of economic growth elsewhere, agriculture's share of GDP will fall unless growth in this economy is biased in favour
of agriculture (unlike for the world economy as a whole). The pro-agricultural
bias in its growth (or in its price distortions - see below, "some qualifications") would have to be greater for agriculture's share of national product not
to decline, the larger the fall in the relative price of food internationally. And
third, the pattern ofdomestic consumption in this model with only two tradable goods is not relevap.t in determining the production pattern in the small
open economy, again in contrast to the situation for the world economy as a
whole; however, as will become clear below, domestic consumption does affect
the trade pattern.
In reality, a large part of each economy involves the production and con-
201
sumption of non-tradable goods and services ( N). These are items for which
the costs of overcoming barriers to trading internationally (especially transport costs) are prohibitively expensive. If it can be shown that the share of
tradables in GDP declines with economic growth, then this increases the likelihood even more that agriculture's share of GDP will decline in growing
economies.
To examine this possibility, combine the F and M sectors into a super-sector
of tradables, T. The price ofT for our small open economy is given by the
international prices ofF and M. The price of non-tradables (N), however, is
determined by domestic demand and supply conditions because, in equilibrium, domestic demand for non-tradables has to equal domestic supply.
Given that the non-tradables sector is roughly equivalent to the services
sector (some services are tradable but some goods are non-tradable), and that
services account for about one-third of expenditure in low-income countries
and two-thirds in high-income countries, it is reasonable to assume that the
income elasticity of demand for non-tradables exceeds one, and hence the corresponding elasticity for all tradables (including food) is less than one. [ Lluch
et al. (1977, table 3.12) obtained estimates of expenditure elasticities in the
range 1.3-1.8 for major groups of services. More recent cross-sectional studies
support the expectation of an income elasticity of demand for services of well
above unity for low-income countries but gradually approaching unity as countries get richer ( Kravis et al., 1983; Summers, 1985).]
Figure 1 can be used to illustrate what happens when a small open economy
with a tradables sector grows, if on the axes F is replaced by T and M by N.
Then E is the initial equilibrium point (when demand for and supply of nontradables are equated) and price line 1 represents the initial price of nontradables relative to tradables. Now assume the price ofT remains unchanged
for the moment but there is economic growth which shifts out the production
possibility curve. If the curve moves out equiproportionately from AB to A' B'
then necessarily both the price and quantity of non-tradables relative to tradables must increase (a real exchange rate appreciation), given that E' must
be to the southeast of the ray OE extended because of the income effect. That
is, tradables contribute a declining proportion of GDP in an economy whose
productivity growth is not sectorally biased but whose consumption growth
pattern is biased against tradables.
As with the earlier discussion of the global economy without non-tradables,
there are two extreme sets of circumstances in which, conceptually, the nontradables sector of a growing economy would not expand its share of GDP. One
is represented by point E", the other by E "'. To reach E", growth would have
to be heavily biased in favour of non-tradables production and the elasticity of
substitution in consumption between T and N low for the relative price of nontradables to fall sufficiently to offset the increase in production of non-tradables relative to tradables. Alternatively, to reach E"' growth would have to be
202
heavily biased in favour of tradables production and again the elasticity of
substitution in consumption between T and N low for the relative price ofnontradables to rise sufficiently to offset the decrease in production of non-tradables relative to tradables. The first possibility seems most unlikely, given that
productivity growth has, if anything, been biased against the non-tradables
sector (Clark, 1957). The second possibility is less remote, but recall that it
requires an extreme bias in productivity growth against non-tradables to offset
the consumption bias in favour of tradables: it is not sufficient, for example,
to be only biased enough for point E 0 to be attained.
Thus, since agriculture's share ofthe tradables part ofGDP is likely to decline,
and the tradables' share of GDP is also likely to fall, introducing a non-tradables sector into the model to make it more closely represent a real economy
makes it even more likely that agriculture's share of GDP will decline with
economic growth.
Employment share
The above reasoning is also sufficient for explaining the decline in agriculture's share of employment, except in cases where agricultural labour productivity grows extremely slowly relative to labour productivity in other sectors.
Where labour productivity is growing more rapidly on farms than in non-farm
jobs, as appears to have been the case in industrial countries during recent
decades (Table 1) , there is even more reason to expect agriculture to become
a less important employer. In fact, the number of people employed on farms in
industrial countries has declined absolutely, the rate of decline averaging 2.8%
per year during the period 1960-1981 (Organisation for Economic Cooperation and Development, 1983).
Export share
Why is agriculture's share of exports less strongly correlated with per capita
income than is agriculture's share of GDP or employment, as shown in the
above regression equations? Consider again a small economy within an open
global economy in which the international price of food relative to other tradables is declining over time because productivity growth in the rest of the world
is not strongly enough biased against agriculture to offset the effect of a relative
decline in food demand. If this small economy is not growing, then clearly the
share of agriculture in exports will decline: the price change would discourage
domestic food production and encourage domestic food consumption, while the
opposite would occur in the domestic market for non-foods. If this small economy is growing and if its productivity growth is sectorally unbiased, then this
tendency for agriculture's share of exports to fall would be weakened because
domestic demand would grow less rapidly for food than for non-food. And it
203
would be weakened further if productivity growth grew faster in agriculture
than non-agriculture in this small economy.
Thus it is understandable that the share of agriculture in exports is less
strongly (negatively) correlated with per capita income than agriculture's
shares of GDP and employment. A negative relationship nonetheless is likely,
both for economies which are growing only slowly and, especially, for economies where productivity growth is fastest in the industrial sector; a positive
relationship will occur only for those economies where productivity growth in
agriculture is sufficiently greater than in manufacturing so as to be able to
generate additional food export volumes to more than offset the long-term
decline in the real international price of food. Note, however, that a declining
contribution to the country's exports from agriculture does not necessarily imply
a declining agricultural comparative advantage as revealed by the pattern of
its export specialisation. This is because agriculture's importance in world trade
may be falling even faster than its importance in this country's exports.
Some qualifications
The above discussion has abstracted from the fact that many economies
seriously distort incentives faced by producers and consumers. In particular,
it is common for low-income economies to tax agriculture relative to manufacturing, for high-income (especially food-deficit) economies to subsidise agriculture relative to manufacturing, and for growing economies to gradually
change from the former to the latter as they industrialise. [See, for example,
Little et al. (1970), Balassa et al. (1971), Anderson and Hayami (1986) and
Tyers and Anderson (in press). The latter two studies also suggest reasons for
the growth of agricultural protection in industrialising countries.] Any one
economy following this pattern after having taxed agriculture would slow the
pace of decline of its agricultural sector, both because of the direct effect of
price changes on sectoral output levels and on their changing per unit valuation at distorted domestic prices, and because ofthe indirect effect those changes
have in boosting investments in new farm technology. Indeed the latter can be
large enough to actually reverse the economy's net food trade position, as has
been the case in Western Europe and China during recent years. When many
economies conform to this distortions pattern simultaneously, the effect on
international food prices depends on the proportion of world food markets that
is low priced as compared with the proportion that is high priced. Over time,
however, as each economy industrialises and so moves towards the agricultural
protection camp, the net effect will be to add to the downward pressure on the
international price of food relative to other tradables. As a result, the agricultural sector of undistorted economies will be under pressure to decline even
faster than would be the case in a distortion-free world.
The above models have also ignored the fact that intermediate goods and
204
TABLE2
Value added share (%) of agricultural (manufacturing) production measured at domestic prices,
Industrial Market Economies, 1963-1983"
Australia
Austria
Canada
EC-IOb
Japan
Korea
New Zealand
Norway
Sweden
Taiwan
United States
Weighted averagec
1963-65
1970-73
38 (41)
33
70
38
57
62
80
63
52
54
56
43
49
43
65 (39)
(33)
59
55
64
42
48
(40)
(39)
(44)
(46)
(40)
(40)
(40)
(33)
(41)
(34)
(39)
(38)
(37)
(38)
(46)
(41)
1975-78
33
68
29
51
62
77
53
52
53
54
37
43
(42)
(37)
(33)
(40)
(32)
(36)
(33)
(31)
(37)
(43)
(41)
1980-83
30
66
22
46
54
67
45
50
48
50
34
39
(38)
(35)
(33)
(39)
(30)
(33)
(33)
(28)
(34)
(42)
(38)
"Manufacturing sector shares are shown in parentheses; in the final column they refer only to
1980. Because of different statistical methods used, shares are not as comparable across countries
as they are over time for each country.
bWeighted average for France, Federal Republic of Germany and the United Kingdom for manufacturing shares, with weights based on 1980 value of manufacturing production.
cweights based on 1980 value of agricultural (manufacturing) production at domestic prices.
Source: Manufacturing data are from United Nations (1985a, b and earlier issues). Agricultural
data for Austria, Japan, Norway and Sweden are from United Nations (1985a and earlier issues),
for the European Community are from Eurostat (1985 and earlier issues) and for other countries
are from the following national sources: Australian Bureau of Agricultural Economics (1986),
Agriculture Canada (1980 and 1984), Korean Ministry of Agriculture and Fisheries (1981 and
1985), New Zealand Department of Statistics (1985 and earlier issues), Council for Economic
Planning and Development (1985), Department of Agriculture and Forestry (1985 and earlier
issues) and U.S. Department of Agriculture (1984 and earlier issues).
services are used in the production process. To include them requires two modifications. One is to recognise that agriculture's share of GDP then refers to a
percentage of value added rather than of total value of production. This would
not matter if the value-added share of output was constant or changed at the
same rate in each sector. But, as Table 2 shows, that share has been declining
much more rapidly in agriculture than in manufacturing: during the two decades to 1983, the share for industrial countries declined from 48 to 39% for
agriculture but only from 40 to 38% for manufacturing, measured at domestic
prices. [The relative decline in agriculture's share would have been even greater
if measured at international prices, given the gradual growth in agricultural
protection mentioned above.] This adds to the reasons for the decline in agriculture's share of GDP. The second modification is to consider the sectoral
source of intermediate goods and services. Casual empiricism suggests the nonfarm sectors traditionally have supplied a disproportionately large and
205
increasing share of those inputs, so this again strengthens the expectation of
a decline in agriculture's share of GDP.
Throughout, the discussion has used the term food to describe agricultural
output. While non-food agricultural products are slightly different in that the
income elasticity of demand for them is not necessarily below one and falling,
they are relatively small contributors to total agricultural output, so the conclusions are unlikely to be altered if they had been specifically included in the
analysis.
Conclusion
This analysis allows the following conclusions to be drawn. First, in a growing world economy (or a closed national economy), agriculture's shares of
GDP and employment are likely to decline because the income elasticity of
demand for food is less than one; to avoid this would require a heavy bias in
productivity growth towards the non-farm sector. A corollary is that the world
price of food relative to non-food is likely to decline with global economic
growth.
Second, in a small open economy faced with a decline in the relative price of
food because of economic growth elsewhere, agriculture's shares of GDP,
employment and exports are likely to decline. This is so even if this small
economy is itself not growing, because of the decline in international food prices.
If it is growing, then agriculture's shares of GDP and employment, if not of
exports, are even more likely to decline because of the low domestic income
elasticity of demand for food. To avoid declining requires either a strong bias
in productivity growth towards the farm sector (in contrast with the global or
closed economy case) and/ or a heavy distortion of incentives towards agriculture. Experience suggests that even then the decline in agriculture's shares of
GDP and employment are unlikely to be arrested: Western Europe and China
provide recent examples of reversing the downward trend in agriculture's share
of exports, but not GDP and employment, through price policy changes which
improved incentives for farmers.
Third, the fact that many countries tend to distort incentives increasingly
in favour of farmers as their economies industrialise, which further depresses
the real international price of food, only adds to the likelihood of agriculture
declining in undistorted economies.
Finally, it should be stressed that it does not follow from this analysis that
farmers necessarily lose from economic growth. Part of the reason for declining
real food prices is increased productivity on farms, and part of the reason for
labour moving out of agriculture is improved income-earning prospects in nonfarm sectors. Thus farm households may well be sharing in the fruits of economic growth either through new farm technologies which raise farm labour
and land productivity and/or through household members taking higher-pay-
206
ing jobs in other sectors. Nor does it follow that policies should be adopted to
accelerate resource movements away from the declining agricultural sector. On
the contrary, the policy response advocated by Prebisch (1964) and others to
declining terms of trade for countries exporting primary products, namely, protecting the import-competing manufacturing sector so as to divert resources
away from primary production, would only make things worse for farmers and
the total economy, as it would worsen the domestic terms of trade for nonmanufacturing sectors as well as reduce national income and hence non-farm
job prospects for farm household members.
References
Agriculture Canada, 1980, 1984. Selected Agricultural Statistics for Canada. Ottawa.
Anderson, K. and Hayami, Y., 1986. The Political Economy of Agricultural Protection: East Asia
in International Perspective. Allen and Unwin, Sydney.
Australian Bureau of Agricultural Economics, 1986. Commodity Statistical Bulletin. Australian
Government Publishing Service, Canberra.
Balassa, B. and Associates, 1971. The Structure of Protection in Developing Countries. Johns
Hopkins University Press, Baltimore, MD.
Clark, C., 1957. The Conditions of Economic Progress. Macmillan, London, 3rd edn.
Council for Economic Planning and Development, 1985. Taiwan Statistical Data Book. Taipei.
Department of Agriculture and Forestry, 1985. Taiwan Agricultural Yearbook. Taipei.
Eurostat, 1985. Yearbook of Agricultural Statistics. Brussels.
Grilli, E.R. and Yang, M.C., 1986. Long-term Movements of Non-fuel Primary Commodity Prices:
1900-83. Economic Analysis and Projections Department, The World Bank, Washington, DC
(mimeograph) .
Johnson, D.G., 1973. World Agriculture in Disarray. Fontana, London.
Korean Ministry of Agriculture and Fisheries, 1981, 1985. Yearbook of Agriculture and Forestry
Statistics. Seoul.
Kravis, I.B., Heston, W. and Summers, R., 1983. The share of services in economic growth. In:
F.G. Adams and B.G. Hickman (Editors), Global Econometrics. MIT Press, Cambridge, MA.
Kuznets, S.S., 1966. Modern Economic Growth: Rate, Structure and Spread. Yale University
Press, New Haven, CT.
Little, I., Scitovsky, T. and Scott, M., 1970. Industry and Trade in Some Developing Countries:
A Comparative Study. Oxford University Press (for OECD), London.
Lluch, C., Powell, A.A. and Williams, R.A., 1977. Patterns in Household Demand and Savings.
Oxford University Press, New York.
New Zealand Department of Statistics, 1985. Official Year Book. Wellington.
Organisation for Economic Cooperation and Development, 1983. Historical Statistics 1960-1981.
OECD, Paris.
Prebisch, R., 1964. Towards a New Trade Policy for Development. United Nations, New York.
Rybczynski, T.M., 1955. Factor endowments and relative commodity prices. Economica, 22:
336-341.
Sapsford, D., 1985. The statistical debate on the new barter terms oftrade between primary commodities and manufactures: A comment and some additional evidence. Econ. J., 95: 781-788.
Schultz, T.W., 1945. Agriculture in an Unstable Economy. McGraw-Hill, New York.
207
Spraos, J., 1980. The statistical debate on the net barter terms of trade between primary commodities and manufactures. Econ. J., 90: 107-128.
Summers, R., 1985. Services in the International Economy. In: R.P. Inman (Editor), Managing
the Service Economy. Cambridge University Press, New York, pp. 27-48.
Tyers, R. and Anderson, K., (in press). Distortions in World Food Markets. Cambridge University Press, Cambridge.
United Nations, 1985a. National Accounts Statistics: Main Aggregates and Detailed Tables. United
Nations, New York.
United Nations, 1985b. Statistical Yearbook. United Nations, New York.
U.S. Department of Agriculture, 1984. Economic Indicators of the Farm Sector. USDA, Washington, DC.
World Bank, 1983, 1984, 1985. World Development Report. Oxford University Press, New York.